ST-70A Modifications Package

A set of design enhancements for the Eico ST-70 integrated amplifier

Rev.5

Copyright © 2010-2018 by Stephen H. Lafferty. All rights reserved.

WARNING!Voltages used in the Eico ST-70 may be LETHAL. Always make sure that the power is discon-nected and discharge each power supply cap AND C37 and C38 before wiring in the unit. Watchout for coupling caps too. When testing inside the unit while it is operating, BE EXTREMELYCAREFUL. Use only one hand when possible. Remove jewelry. HV DC can contract muscles andprevent you from releasing grasp of a HV terminal. Dead techs aren’t much fun.

Contents

Introduction .................................................................................................................................... 3Personal note............................................................................................................................. 3Acknowledgment ...................................................................................................................... 3

Overview ........................................................................................................................................ 3Applicability to other products ....................................................................................................... 4Detailed modifications .................................................................................................................... 4

Phono preamp ........................................................................................................................... 4Changes .............................................................................................................................. 4Adjustments ........................................................................................................................ 4Fixed resistor alternative .................................................................................................... 5

Line stage .................................................................................................................................. 5Changes, Channel-1 ............................................................................................................ 5Changes, Channel-2 ............................................................................................................ 5Physical layout .................................................................................................................... 5Checking C31 ..................................................................................................................... 6

Power amplifier ........................................................................................................................ 6Changes, Channel-1 ............................................................................................................ 6Changes, Channel-2 ............................................................................................................ 7Adjustments ........................................................................................................................ 8

Testing the mods ............................................................................................................................. 8Phono preamp ........................................................................................................................... 8Line stage .................................................................................................................................. 9Power amplifier ........................................................................................................................ 9

Output power ...................................................................................................................... 9Distortion at 10kHz ............................................................................................................ 9

Before and after performance ......................................................................................................... 9Output power ............................................................................................................................ 9Line-level signal to noise ratio ................................................................................................. 9Distortion ................................................................................................................................ 10Stability at high frequency ...................................................................................................... 10Stability at low frequency ....................................................................................................... 10RIAA phono accuracy ............................................................................................................. 11Frequency response ................................................................................................................ 11

Parts list ........................................................................................................................................ 11Complete schematic of the ST-70A .............................................................................................. 12

3

ST-70A Modifications Package

IntroductionThis describes a set of modifications (hereafter “mods”) to the Eico ST-70 stereo integratedamplifier, which I will call the “ST-70A”. The upgrade fixes problems with line-level noise,phono preamp bass response, increases low frequency and high frequency power output, reducesdistortion and improves power amplifier stability into difficult loads. After listening to the ST-70A for many hours now, I can say that the project has been an audible and gratifying success.

The original ST-70 kit was bought by my Dad in August, 1962 and was assembled by my bigbrother, Bill. I remember watching in fascination as all the pretty-colored wires and componentsbecame an awesome music-maker. The ST-70 and other kits were part of the inspiration whichled me to become an electronic design engineer. Little did I know, as I watched at the age of ten,that I would be redesigning it 46-years later! Calling it a “redesign” is too strong a term, though. Thegoal has been to do the maximum good while making minimum changes to the original design.

Important notes:• Component names for Channel-1 will be referenced to represent both channels.• The individual mods are: Phono preamp, driver balance adjust and line amp/power amp.

You can do the mods singly but the line amp and power amp mods must be done together.• The line amp mod decreases the current used by those stages. This results in a beneficial

increase in voltage at power supply nodes W, Y and Z.

OverviewThere were four motivations for the mods: (1) Too much hiss noise in the line stage. (2) Substan-tial bass loss in the phono preamp. (3) Lack of a grounded common at the speaker outputs whichcaused problems for connections to speaker switches, headphones and instruments. (4) Anopportunity to increase power and decrease distortion. There is no change to the original inputsensitivity for either the phono or the line inputs. The mods address three areas of the unit:

• Phono preamp positive feedback adjustment. Eico used positive feedback to achieveadequate low frequency gain in the phono stage. Unfortunately, it is sensitive to tubetransconductance. Either Eico used higher gain tubes than we have now or they decidedto include a free rumble filter. I measured the preamp about 10dB down at 20Hz and 4dBat 35Hz. The adjustment corrects that to 0.7dB at 20Hz and 0.0dB at 35Hz.

• Line stage noise reduction and gain increase. The presence of R41 without a stable inputresistor results in high noise at low settings of the level control. Also, level in the tonestage is unnecessarily low due to excessive gain in the power amp. Changes increase thegain of V3A by about 10dB and eliminate the noise enhancement there.

• Power amp (hereafter “PA”) gain reduction, distortion reduction, stability improvementand common speaker grounding.

- The ground was moved from the 4Ω tap to the common speaker tap and this wastaken into account in the new feedback and compensation components.

- The original PA sensitivity is about 0.2Vrms for full output. This is changed to about0.65Vrms to reduce the effect of hum and noise from V3 and V5.

- Additional loop compensation in the PA permitted the resulting 10dB increase in loopgain. This led to a 3X reduction in PA distortion which in turn increased low frequency

4

and high frequency power output (measured at 1% distortion).- Moving feedback from the 16Ω to the 8Ω tap reduced distortion further at that tap.- Adding AC Balance pots reduces distortion. It greatly reduces its dependence on

R61 and R63 and on tube matching (V11/9, V7A/B).- The new loop compensation improved stability dramatically, allowing the amp to

remain stable with pure capacitive loads. Some believe thatexceptionally solid high frequency stability leads to audibleimprovements in sound.

- Changes also improved low frequency stability andreduced overload recovery time.

Applicability to other productsThe Eico ST-40 integrated amp and the ST-84 preamp both use thesame phono preamp design as the ST-70, so that part of the modcan be used for those. Since the ST-40’s line stage and power ampare different, the rest of the mod is not applicable.

Detailed modificationsNote: All resistors can be 1/4-watt or larger carbon film unlessotherwise specified. I used 1/2W when I had it, for visual harmony.

Phono preampIn the schematic at right, increasing the coupling caps extends thebass response but more open loop gain is needed to support theRIAA curve. Reducing R27 with the pot boosts positive feedback,increasing open loop gain and allows adjustment to accommodatetube variations.

ChangesC7 from 0.025µF to 0.1µF, 400V.C8 from 0.025µF to 0.1µF 400V.C9 from 0.025µF to 0.1µF 400V.C10 from 0.025µF to 0.1µF 400V.Bridge R27 with a 500K pot (15-turn preferred).Bridge R28 with a 500K pot (15-turn preferred).

As shown in the photo of the bottom of V1 at right, Imounted the pot by super-gluing its back end to the chassis.It is also braced against the terminal strip and its top lead issoldered to the adjacent terminal.

AdjustmentsPreset the pot for highest resistance.Apply 1kHz 10mVrms to phono input and read level at tape output (1.14Vrms nominal).Change frequency to 35Hz and the reduce the generator level by 18.2dB (to 1.23mVrms).Adjust R27A for the same level as before at the tape output.

5

V3

Level

V4Balance

R41

R33A

C15xtra-C

Repeat for the other channel.Adjust hum balance for both channels.

Fixed resistor alternativeIf you are unable to make this kind of adjustment, you may be able toget adequate performance by bridging R27 with a fixed resistor. Thevalues of the pots in my ST-70 were about 420K and 290K. I wouldrecommend using 360K, based on this. In the ST-84 I worked on years ago, Isettled on 200K fixed resistors. Though it’s the same circuit, the plate voltagesthere are roughly half those in the ST-70, accounting for the difference.

Line stageAdding R33A stabilizes the gain of the stage at low Level-control settings,reducing hiss and hum. R41 is increased to increase level in the tone amp andat the power amp, reducing hum and hiss. The tube is changed and cathodebypassed to maximize open loop gain and support the increase in stage gain.

Changes (shown at right), Channel-1Change V3 from 12DW7 to 12AX7.Add R33A, 100K between wiper of Channel-1 Level control andV1-2. (Channel-1 section is closest to front panel.)Change R37 from 33K to 100K.Change R41 from 220K to 1M.Change R43 from 1.5K to 2.2K.Add C43A, 100µF aluminum electrolytic. 6V or higher.

Changes, Channel-2Change V4 from 12DW7 to 12AX7.Add R34A, 100K between wiper of Channel-2

Level control and V1-2. (Channel-2 section is farthestfrom the front panel.)

Change R38 from33K to 100K.

Change R42 from220K to 1M.

Change R44 from1.5K to 2.2K.

Add C44A, 100µFaluminum electrolytic.

Physical layoutIn the original layout atright, the grid of V3 and R41 tie to the wiper of theLevel control. We need to insert R33A between the gridand the wiper, so the new value of R41 must go to thegrid pin instead the wiper, as shown at left. Extend onelead of R33A to reach the wiper of the control. R41 is

also extended to reach the terminal of C15. R41 and R33A should be as close as possible to V3-2.That node is sensitive to hum pickup and must be kept small.

V3

Level

R41

6

Checking C31In my unit, C31 and C32 were reading0.012µF instead of the nominal0.015µF. This was causing rolloff ofperhaps 1-2dB in the 20kHz region. Tocheck it with a cap meter you will needto disconnect one side from R87. Nothaving a suitable replacement, I aug-mented the caps with an additional3.3nF shown as “xtra-C” in the photoon the previous page. It’s a good ideato check the values of all componentswhich can be easily read. My rule is toreplace when the value is out of toler-ance. You may notice in the photo thatR50, R87, R88 have been replaced.

Power amplifierAs shown at right, the speaker outputsnow have common connected toground and the center speaker featureis no longer supported. Moving thefeedback to the 8Ω output tap reducesdistortion for that tap. Feedback resis-tors R85/56 are changed to reduceclosed loop gain, complementing the increase in line stage gain. That also increases feedbackfactor, reducing distortion and increasing power at a specified distortion. To maintain and im-prove low frequency stability, C19/21 are decreased, feedback is moved to the other side of R55and C33 is increased. The decrease in C19/21 does NOT impair bass performance, due to theloop gain. C37 is increased to support the low frequency loop gain. R53A/59A and C61A/59Aare added to introduce pole-splitting frequency compensation, keeping and improving highfrequency stability with the higher loop gain. Finally, AC balance control R61A is added toreduce distortion due to tube and resistor imbalances.Changes, Channel-1:

Drill holes* and mount the new AC balance pots R61A, R62A in the chassisnear the existing DC balance pots R79/80. You can see the locations in thephotos of that section on the next page.Remove the ground connection from the 4Ω terminals. The Channel-1terminal board is shown at right. Make a sturdy connection between groundand the common terminal (where the black lead from the output transformeris).Change R85 from 6K to 4.3K and connect it to the 8Ω terminal.C35 leads are too short to reach the 8Ω terminal. Replace it with 120pF.

*Since this is steel, recommended drilling tips include: start with a center punch, use cutting oil, use sharp,coated drill bits, use a drill press, set the correct speed and start with a pilot hole.

C

16

4

8

R85C35

7

Refer to the photo at right:Move the feedback wire to the other sideof R55. (The wire runs from R85/C35.)Change C33 from 25µF to 470µF.Change R56 from 100Ω to 150Ω.Change C19 and C21 from 0.1µF to0.047µF. (May have to extend leads.)Change C37 from 0.25µF to 1µF 250V ormore electrolytic. Note polarity. You cansolder the ground side to a lug on the V7socket.Add the two RC networks R53A/C61Aand R59A/C59A, as shown in the detailat left. Note the heatshrink at the junc-tion. Since this is sensitive to straycapacitance, do not try to connect thejunction to a terminal strip.

Remove the existing R61 and R63. Thesupply terminal used by R61 is no longerneeded. Remove the red wire feeding it.Connect new pot R61A wiper to powersupply point “U” at the terminal abandoned by R63.Install the new 1W resistor values: R61 changes from 28.75K to 22K, R63 from 33K to27K. Note that modern 1W resistors are smaller than the ones found in the unit.

Changes, Channel-2:Remove the ground connection from the 4Ωterminal of the Channel-2 terminal board.Make a sturdy connection between groundand the common terminal, where the blacklead from the output transformer is.Change R86 from 6K to 4.3K and connect itto the 8Ω terminal.C36 leads are too short to reach the 8Ωterminal. You will need to replace it with120pF.

Refer to the photo at right:Move the feedback wire to the other side ofR56. (The wire runs from R86/C36.)Change C34 from 25µF to 470µF.Change R57 from 100Ω to 150Ω.Change C20 and C22 from 0.1µF to0.047µF. (May have to extend leads.)Change C38 from 0.25µF to 1µF 250V ormore electrolytic. Note polarity. You cansolder the ground side to a lug on the V8socket.

R62A

unusednow

V6

V8

V10

“U”Po

int “

U” fe

ed to

Cha

n-1

Chan-2 feedback

V5

V7

V11

V9

R61A

Chan-1 feedback

8

Add the two RC networks R54A/C62A and R60A/C60A. See detail on the previous page.Note the heatshrink at the junction. Since this is sensitive to stray capacitance, do not tryto connect the junction to a terminal strip.Remove the existing R62 and R64. The supply terminal used by R62 is no longer needed.Remove the red wire feeding it. Since Channel-1 was fed from this terminal, re-route itsfeed directly to the cap marked “U” in the photo. (Since my hookup wire is rated at 300Vand “U” is about 350V, I spliced the original Eico wire. Notice the (double) heatshrink.Connect new pot R62A wiper to power supply point “U” at the big cap C42A.Install the new 1W resistor values: R62 changes from 28.75K to 22K, R64 from 33K to27K. Note that modern 1W resistors are smaller than the ones found in the unit.

Adjustments:Preset the new AC Balance pots R61A, R62A to approximate center.Connect 8Ω dummy loads to both channels.Allow plenty of time for warmup and make sure DC bias and balance is set properly.Apply 1kHz, 0.5Vrms to a Channel-1 line level input and set Level for 16.5Vrms out.Make sure it is comfortably away from clipping. Reduce level if necessary.Adjust R61A for minimum distortion. Alternative: If you cannot measure distortion,apply the signal as above and connect a DC voltmeter between the bias test points. AdjustR61A for zero volts (±4mV).Repeat for Channel-2.

Testing the modsThese are pretty much minimal functionality tests to verify construction. Connect a dummy loadto the speaker terminals of both channels. Required test equipment:

• Audio generator• AC voltmeter and/or distortion analyzer• Oscilloscope

Phono preampThis was basically tested with the adjustment of R27a. If you would also like to check the fre-quency response of the phono preamp:

• Apply 1kHz, 10mVrms to the phono-B input. (Phono-B has the correct cartridge load.)• Read reference level at the tape output (1.14Vrms nominal).• Apply the following frequencies and levels. Read tape output levels. Deviations from

reference level are response errors. Freq Rel. Level Abs. Level 20Hz -19.3dB 1.09mV 50 -16.9 1.42 100 -13.1 2.22 200 - 8.2 3.88 500 - 2.6 7.37 1K 0.0 10.0 2K + 2.6 13.5 5K + 8.2 25.7 10K +13.7 48.6 20K +19.6 95.7

9

Line stageTo test the line stage:

• Apply 1kHz, 350mVrms to a line-level input and set the Selector to it.• Set the Level control to maximum.• Measure the output of the line stage at the terminal where R41 and C15 join.• It should read approximately 0.84Vrms.• Verify that the waveform looks clean on an oscilloscope.• If you wish to measure distortion, I read 0.07% there.• Repeat for the other channel.

Power amplifierOutput power

• Apply 1kHz, 350mVrms to a line-level input and set the Selector to it.• Set the Level control to maximum; Bass and Treble controls at center.• Find the generator level at which clipping occurs. It should be approximately 350mVrms.• Set the output voltage just below clipping• Measure the output across the dummy load. It should be about (35W): (choose a load)

- 11.8Vrms for a 4Ω load- 16.7Vrms for a 8Ω load- 23.7Vrms for a 16Ω load

Distortion at 10kHz• Set up as above at 10kHz. Adjust Level for 8Vrms output.• Measure distortion at the output. I measured 0.14%.

That completes the basic checkout. Enjoy your new amplifier! If you have problems, you can reachme at http://www.tronola.com. Post in the article reader comments or see “About” on the home page.

Before and after performanceThis section presents some of the measurements I made before and after the mods. The referenceline voltage was 121Vrms.

Output power (One channel, PA only, 8ΩΩΩΩΩ load, 1% distortion)Freq Before After20Hz 13W 19W30Hz 30 331kHz 32 3520kHz 10 21

Power at 20Hz increased 46% due to the reduction in distortion from the increased PA loop gain.Power at 20kHz more than doubled for the same reason.

Line-level signal to noise ratio (Full ST-70, at speaker output, 8ΩΩΩΩΩ load, relative to 35W) LevelFreq Range Control Before After400Hz-30kHz min 70dB 82dB400Hz-30kHz max 76 8220Hz-30kHz min 64 7920Hz-30kHz max 61 71

10

The annoying hiss at low settings of the Level control was reduced 12dB and is very quiet in alistening test with efficient speakers. Hiss can be heard 18” from the tweeter. Hum was reduced15dB. (The 71dB figure was limited by the test setup which had to match the “before” testing.)

Distortion (One chan, 8Vrms output, PA only, 8ΩΩΩΩΩ load)Freq Before After20Hz 0.70% 0.11%30Hz 0.43 0.061kHz 0.12 0.035kHz 0.19 0.0610kHz 0.37 0.1320kHz 0.85 0.33

Distortion has been dramatically reduced by the increase inloop gain, particularly at low frequencies. Adding the ACbalance control and changing the feedback tap also helped.This is measured at about half the maximum voltage output.

Stability at high frequency (PA only)High frequency (18kHz) square wave response is shownbefore (top right) and after (right) the mods. Transientresponse looks about the same. (The bottom trace is thin-

ner, due to averaging.)Though HF stabilityappears unchanged, thestock ST-70 has been reported to be unstable when loadedsolely by a capacitive load. I found that the worst case load forstability is 0.08µF and was able to achieve stable operationwith that load. Transientresponse with solely a0.1µF load is shown atleft.

Stability at low frequency (PA only)• Very low frequency square wave response is shown atright, before the mods (250ms/div). Note the rebound.• At bottom right after the mods, there is very littlerebound, showing a more stable loop.

Moving feedback to the cathode of V5, along with thechanges in C33/19/21 really helped low frequency stability.This can be important because it affects how fast the amprecovers from an overload. Reducing C19/21 also helpedrecovery time by cutting the output bias time constant inhalf. In listening tests, there was no discernable recoveryartifact after clipping, either audibly or on a scope trace.

11

RIAA phono accuracy• Before the mods, I measured phono response

about 10dB down at 20Hz and4dB at 35Hz.

• After the mods, 0.7dB at 20Hz and 0.0dB at35Hz. Final phono response (measured atTape Out) is plotted at right.

Phono preamp response is within ±0.4dB over allbut the lower extremity of the range. The adjustmentnot only nailed the problem, it allows for correctionsas the tubes age.

Frequency response (PA only)Item Before After _20Hz resp -0.02dB -0.14dB20kHz resp -0.13dB +0.10dBLF -3dB freq ~3Hz 2.6HzHF -3dB freq 150kHz 138kHzLF -0.5dB freq na 16Hz

Although the response of the PA appears a fraction worse at the bottom end of the audio range, itis certainly not audible. The difference can be attributed to the fact that the subsonic response nolonger peaks, indicating better LF stability now.

Parts listQty Reference Description Comment4 C19/20/21/22 0.047uF 600V polypropylene capacitor polyester okay2 C33/34 470uF 6V aluminum electrolytic capacitor higher voltage okay2 C35/36 120pF 100V NPO ceramic capacitor higher voltage okay2 C37/38 1uF 450V aluminum electrolytic capacitor 250V okay2 C43A/44A 100uF 6V aluminum electrolytic capacitor higher voltage okay4 C59A/60A/61A/62A 47pF 500V NPO ceramic capacitor higher voltage okay4 C7/8/9/10 0.1uF 400V film capacitor2 R27A/28A 500K 15-turn pot Single-turn okay4 R33A/34A/37/38 100K 1/4W carbon film resistor 1/2W okay for aesthetics2 R41/42 1M 1/4W carbon film resistor 1/2W okay for aesthetics2 R43/44 2.2K 1/4W carbon film resistor 1/2W okay for aesthetics2 R56/57 150ohm 1/4W carbon film resistor 1/2W okay for aesthetics4 R53A/54A/59A/60A 16K 1/4W carbon film resistor 1/2W okay for aesthetics2 R61/62 22K 1W carbon film resistor2 R61A/62A 10K cermet potentiometer wirewound okay. NOT carbon. Note-22 R63/64 27K 1W carbon film resistor2 R85/86 4.3K 1/4W carbon film resistor 1/2W okay for aesthetics2 V3/4 12AX7 vacuum tube

Note-1: New additions have an "A" suffix in the reference name.Note-2: I hastily chose Mouser 652-51AAA-B28-A15L but it costs $8.70.

AcknowledgmentsI would like thank my good friend David Gillespie for his advice and support throughout thiseffort. He suggested the component values which worked so well for the line stage. Thanks alsoto brad347347 on the Eico tube amps Yahoo group for various corrections.

dB

Freq

12

Com

plete schematic

of the ST-70A(page-1 of 2)

13

(page-2 of 2)